Sherrie Rice
December 2024—A National Cancer Institute trial known as MyeloMATCH, rolling out now, aims to improve the acute myeloid leukemia survival rate and relies on 72-hour turnaround times for cytogenetics, FISH, flow cytometry, and next-generation sequencing.
In MyeloMATCH (Molecular Analysis for Therapy Choice), researchers are hoping to find new treatments for AML and myelodysplastic syndrome by rapidly matching patients with a trial that tests a treatment designed to target the mutations detected in the patient blood and bone marrow samples.
A panel of clinicians and information technology and pathology experts meets daily to review the final laboratory results from the previous day and make the trial assignments. “It’s quite an amazing effort to put all this together,” said Jerald Radich, MD, the Kurt Enslein endowed chair, Fred Hutchinson Cancer Center, which is a MyeloMATCH laboratory testing site. He spoke in August in a CAP TODAY webinar made possible by a special educational grant from Thermo Fisher Scientific (available at captodayonline.com).
MyeloMATCH is about “where we have been and where we want to go,” Dr. Radich said. He describes the success seen in chronic myeloid leukemia, with the advent of the molecularly targeted tyrosine kinase inhibitor therapy, and the slow progress in AML as a “tale of two leukemias.” The hope is that MyeloMATCH will bring targeted therapy to AML and, he tells CAP TODAY, “make a similar treatment arc of improvement as seen in CML.”
The rate of new cases of CML from the 1980s to about 2000 remained fairly stable. But death rates, which were stable for a long time, then dropped dramatically, owing to the tyrosine kinase inhibitor Gleevec and now its second-generation drugs. CML is unique in that one molecular lesion, BCR-ABL, drives the disease. It serves as a target for drug therapy and as a marker of disease burden, “so that in almost all of the FDA trials for the second-generation drugs, the BCR-ABL, as determined by peripheral blood mRNA testing, is the endpoint as a sensitive marker of the disease burden,” Dr. Radich said.
The question is whether that kind of precision medicine can be achieved for other diseases, like AML, “where death rates have been stubbornly stable for the past few decades,” he said. In its MyeloMATCH initiative, the NCI is aiming to bring modern molecular methods into modern drug development of small-molecule agents, “to try to see what we saw in CML in AML.”
Compared with solid tumors, he said, “AML is a pretty clean space” in terms of its mutational heterogeneity—about 25 to 30 mutations cover nearly all of the possible mutations seen in AML. “Most patients have on average about three to five mutations, as opposed to solid tumors where you might have 20 mutations,” Dr. Radich said. The association of one kind of mutation with another in AML is not random, and some are common, such as DNMT3 with NPM1. “There’s not an infinite number or variety of mutations or how they coexist with each other. So it’s potentially a space to go after as far as defining disease subsets and finding clinical trials designed for specific mutational targets. Things we can define at diagnosis,” he said.
“If we use those molecular underpinnings and we have trials to specific subtypes with specific agents and we target those,” he continued, “can we make AML the new CML? That’s what the MyeloMATCH is about.”
MyeloMATCH has three main features, one of which is the use of genetically driven protocol assignments, using “cytogenetics for rough risk groups and molecular lesions to define the specific drug trials they’ll go on, ” Dr. Radich says.
Second, MyeloMATCH will be predominantly randomized phase two trials, unlike the NCI’s cooperative group program of the past, which was predominantly phase three trials.
Third, as opposed to the traditional way of using overall survival endpoints, which takes five to 10 years, the MyeloMATCH suite of studies will be predominantly measurable residual disease-driven, with flow cytometry right now being the endpoint. “As opposed to five-, 10-year trials, these will all be trials where we’ll make a decision about whether the trial is a success or not at the end of induction for some trials and at the end of a consolidation regimen for others,” Dr. Radich explained.
So if something that has a strong signal is found, it can then be folded into a phase three trial. “We’re trying to move the bar and bring up new drugs and quickly identify whether they’re going to work or not.”
He describes the five years of work behind developing MyeloMATCH’s essential components as a Herculean effort. One component is the information technology team, which is focused on, among other things, tracking enrollment from local sites and following patients throughout the trials.
“They’ve had to build this from scratch, because with all the U.S. inner groups, we’re going to have hundreds of potential sites that could enroll to these trials across the U.S. We’re going to have to track that enrollment. We’re going to have to track the samples as they go to the centralized labs. We have to integrate all the clinical information and lab data for a treatment assignment. And then we have to follow those patients. That’s been a huge informatic undertaking,” Dr. Radich said.
Another component is the agents team that links pharmaceutical companies and new agents with specific and new drug targets. “They have to find the partners, develop contracts. This is hundreds of hours of work to try to secure pharmaceutical involvement in supporting these trials,” he said.
The laboratory team is developing cytogenetics, FISH, flow cytometry, and next-generation sequencing validated assays, all in the CLIA setting and working with the Food and Drug Administration to have an investigational device exemption so these methods can be used for drugs that are not yet approved. “And the trials team integrates all of these labs and tests into new trial design,” Dr. Radich said.
A newly diagnosed patient with AML or MDS will go to any of the NCI National Clinical Trials Network sites to have multiple samples drawn.
One sample will be sent to the local pathologist who will make the diagnosis of AML or not AML. “The FDA has made it clear that we cannot make a call of AML or MDS just on cytogenetics or molecular data. We have to have the local pathologist say that is what it is,” Dr. Radich said.
Samples will then be sent to various laboratories. Flow cytometry will be done at Children’s Hospital Los Angeles in the laboratory of Brent Wood, MD, PhD. Cytogenetics will be performed in the laboratory of Min Fang, MD, PhD, at Fred Hutchinson Cancer Center, and molecular analysis will be done in the NCI molecular characterization laboratory (known as MoCha) at the Frederick National Laboratory for Cancer Research or at Fred Hutchinson. A large number of samples will be sent to nationwide repositories for future work.
“We have various tiers, and the way to conceptually think about these tiers is it’s based on how much disease people have at the time of diagnosis,” Dr. Radich said. The first trials to open are tier one treatment trials, which offer initial induction therapy for all newly diagnosed patients. Tier two trials treat patients for residual disease after initial therapy.
Reassessment by flow cytometry will be done after consolidation therapy. “If those patients are negative by flow, they keep going on trial. If they’re positive, they will then go to the treatment tier two trials, which are essentially MRD Eraser trials, where you have patients who have residual disease and you build a new trial to attack that residual disease and make them negative,” Dr. Radich explained.
For patients for whom it is appropriate to go on to transplant or for whom the drugs fail at tier two, there will be a suite of transplantation and cellular therapy trials in tier three. In tier four trials, highly sensitive next-generation sequencing techniques will be used, mostly duplex sequencing techniques that can potentially detect variant allele frequencies down to one in a million and detect whether there is leftover disease burden at that point. “And then attack those with specific drugs,” he said.
Rolling out now is tier one. Tiers two, three, and four will follow.
For purposes of both patient care and study design, the laboratories’ clinical colleagues said “they would like to give the novel therapeutics at the get-go,” Dr. Radich said. “Not have a strategy where you do standard induction therapy and then, based on the molecular results you get back in a few weeks, adding on drug A or drug B, but rather knowing right up front and then initially treating them with the targeted therapeutic agent.”
To do that, rapid assessment is a must. “So we are going to do cytogenetics, FISH, and a myeloid assay all within 72 hours. We have calls every day where a panel of us are on to review the data that’s coming in. Once we get the sample in the door and all the information at the central labs, we’ll make a call within 72 hours of what protocol the patients will go on. That’s kind of revolutionary,” he said.
The molecular laboratories at Hutchinson and MoCha will do the rapid gene sequencing. Duplex sequencing for the integrated studies will be done at TwinStrand Biosciences and likely Hutchinson too. The Hutchinson laboratory will do the more research-oriented work of single-cell omics and RNA sequencing.
The rapid and chips-based Thermo Fisher Ion Torrent Genexus will be used for sequencing.
It provides automated library preparation using Ion AmpliSeq technology, template preparation with isothermal amplification, and rapid semiconductor pH sensor-based sequencing. “That allows us to go, once we get the DNA/RNA available, from putting it in the machine and getting the answer out, for the actual analysis in the machine, about 30 hours. And then we have the prep up front, so we can get everything done in 48 to 72 hours,” Dr. Radich said.
The NCI myeloid next-generation sequencing assay they’re using is for both RNA and DNA. There are 1,600 DNA hotspots and 800 RNA fusion isoforms and exon splice variants. RNA input is 14.25 ng per sample; DNA is 27.75 ng per sample. “You roughly get about 2 million DNA reads and 300,000 to 500,000 RNA reads per sample. And they’re targeting AML, MDS, myeloproliferative neoplasms, and other myeloid neoplasias,” he said.
Requesting an FDA investigational device exemption “required an enormous amount of work,” he said, adding, “We submitted to the FDA 1,400 pages of data.”
In work led by Cecilia Yeung, MD, medical director of the Hutchinson clinical testing labs, and “D. J.” Jiwani, MD, PhD, of MoCha, the laboratory validated 163 unique samples of patients and healthy donors, including bone marrow, cell lines (cancer and normal), and contrived samples. The DNA covered 45 genes, 1,661 hotspots, 1,052 SNPs, and 609 indels. For RNA: 35 driver genes and 779 fusions.
“The long and short of it is the sensitivity of the assay was 98.62 percent at one center, 98.97 percent at the other center,” Dr. Radich said. Specificities were 100 percent at both MoCha and Hutchinson. “Accuracy is fantastic”—greater than 99.99 percent at both sites. “And the positive percent agreement”—100 percent and 99 percent—“and negative percent agreement”—both 100 percent—“are, again, fantastic.”
“We had to prove we could do this on time, on contrived samples and patient samples. We did that, and we passed the bar as far as the FDA quality criteria we have to meet,” he said.
The limit of blank assesses the background noise for the assay. No false-positives were detected in all variant classes at the sample and variant levels.
The limit of reporting identifies the accuracy of reporting true positive variants above the recommended LoR. All samples will be reported at five percent variant allele frequency SNV/indels, FLT3-ITDs (internal tandem duplication) at one percent VAF, and fusions at 100 reads. “And limit of detection—we’re going to be calling at five percent VAF. We can go lower than that, but this is a level we’re comfortable at calling,” Dr. Radich said.
Among the assay’s limitations are a few blacklisted variants—ASXL1, PRPF8, and RUNX1. “These are areas of basically unclear pathology and [they] have a little bit of variation and recalls that make it difficult to interpret,” he explained.
Expected run failures are another limitation. “Most are from problems with uniformity across the various targets or sometimes no template controls being a little higher than you want. We’re at about 7.5 percent run failures, but if we repeat, we still make the 72-hour criteria.”
A third limitation is related to FLT3-ITDs, which he said are difficult to sequence and were tested at five laboratories with different technologies. “For the Thermo Fisher technologies, we missed three very, very low VAFs, and these are VAFs that we wouldn’t call anyway.”
More critical, Dr. Radich said, is the size of the ITD. “For the Thermo Fisher assay, we can detect up to between 21 and 117 base pairs. If you put the cutoff at roughly 120, you will miss cases, probably between four and eight percent of FLT3-ITDs. But the FDA said that was okay; that was the cost of business. They’re not requiring us to do any reflex testing for those people who were FLT3-negative. So that’s just what we’re going to have to deal with with the assay.”
Four trials are up and running now, with more than 150 centers having activated and more activating daily. One trial is the master screening and assessment protocol, “which allows us to do all this testing, and then we have specific clinical trials,” Dr. Radich said. “We think at steady-state we’ll have about 10 trials open at any one time, and we’re hoping to accrue about 2,000 new patients a year.”
Sherrie Rice is editor of CAP TODAY.